AI Policy Scandal: South Africa Withdraws National AI Framework After "Hallucinated" Citations FoundIn a deeply ironic turn of events, the South African government has been forced to withdraw its newly proposed National Artificial Intelligence (AI) Policy following the discovery of fabricated academic references. The draft, intended to establish a regulatory framework for AI ethics, socio-economic impact, and developmental incentives, is now under scrutiny for potentially being generated by the very technology it sought to regulate.
The "Hallucination" Controversy
The scandal broke when eagle-eyed reviewers noticed that at least 6 out of the 67 citations in the policy’s bibliography referred to academic journal articles that simply do not exist. This phenomenon, known in the AI world as "Hallucination," strongly suggests that the authors may have used a Large Language Model (LLM) to draft the policy without proper human oversight.
Accountability and Official Response
Solly Malatsi, South Africa’s Minister of Communications and Digital Technologies, addressed the issue on X (formerly Twitter), stating that the error is "unacceptable." He confirmed that the draft has been withdrawn and that those responsible will face disciplinary action. Malatsi assured the public that a revised version, thoroughly vetted by human experts, will be presented in due course.
This incident is a prime example of the "human-in-the-loop" phenomenon. While the government is drafting legislation to regulate AI ethics, it is allowing AI to "lie" within the legislation itself. This underscores that no matter how intelligent AI is, final human fact-checking remains essential, especially at the national policy level.
South Africa is not the first country to face this problem. Previously, a lawyer in the US used ChatGPT to draft a lawsuit, only for the AI to cite forged court rulings, leading to a conviction. This incident serves as a wake-up call to government agencies worldwide rushing to implement AI policies, reminding them that "speed" should not be sacrificed for "accuracy."
This raises crucial questions about government transparency: Are AI being used in other areas of public administration? If such complex policies can be written by AI, what about crucial data insights and decisions that impact citizens? This presents a golden opportunity for the South African government to turn this crisis into an opportunity by establishing a standard of "AI Transparency" in its revised draft legislation, setting an example for the world to follow.
Instagram New Algorithm Reach Penalties for Reposters Big Wins for Original Creators.
Source: The Independent
AI Policy Scandal: South Africa Withdraws National AI Framework After "Hallucinated" Citations FoundIn a deeply ironic turn of events, the South African government has been forced to withdraw its newly proposed National Artificial Intelligence (AI) Policy following the discovery of fabricated academic references. The draft, intended to establish a regulatory framework for AI ethics, socio-economic impact, and developmental incentives, is now under scrutiny for potentially being generated by the very technology it sought to regulate.
The "Hallucination" Controversy
The scandal broke when eagle-eyed reviewers noticed that at least 6 out of the 67 citations in the policy’s bibliography referred to academic journal articles that simply do not exist. This phenomenon, known in the AI world as "Hallucination," strongly suggests that the authors may have used a Large Language Model (LLM) to draft the policy without proper human oversight.
Accountability and Official Response
Solly Malatsi, South Africa’s Minister of Communications and Digital Technologies, addressed the issue on X (formerly Twitter), stating that the error is "unacceptable." He confirmed that the draft has been withdrawn and that those responsible will face disciplinary action. Malatsi assured the public that a revised version, thoroughly vetted by human experts, will be presented in due course.
This incident is a prime example of the "human-in-the-loop" phenomenon. While the government is drafting legislation to regulate AI ethics, it is allowing AI to "lie" within the legislation itself. This underscores that no matter how intelligent AI is, final human fact-checking remains essential, especially at the national policy level.
South Africa is not the first country to face this problem. Previously, a lawyer in the US used ChatGPT to draft a lawsuit, only for the AI to cite forged court rulings, leading to a conviction. This incident serves as a wake-up call to government agencies worldwide rushing to implement AI policies, reminding them that "speed" should not be sacrificed for "accuracy."
This raises crucial questions about government transparency: Are AI being used in other areas of public administration? If such complex policies can be written by AI, what about crucial data insights and decisions that impact citizens? This presents a golden opportunity for the South African government to turn this crisis into an opportunity by establishing a standard of "AI Transparency" in its revised draft legislation, setting an example for the world to follow.
Instagram New Algorithm Reach Penalties for Reposters Big Wins for Original Creators.
Source: The Independent
Comments
Post a Comment